Development of Aluminum LEKIDs for Balloon-Borne Far-IR Spectroscopy
S. Hailey-Dunsheath, A. C. M. Barlis, J. E. Aguirre, C. M. Bradford, J. G. Redford, T. S. Billings, H. G. LeDuc, C. M. McKenney, M. I. Hollister
JJournal of Low Temperature Physics manuscript No. (will be inserted by the editor)
Development of Aluminum LEKIDs for Balloon-Borne Far-IRSpectroscopy
S. Hailey-Dunsheath · A. C. M. Barlis · J. E. Aguirre · C. M. Bradford · J. G. Redford · T. S. Billings · H. G. LeDuc · C. M. McKenney · M. I. Hollister the date of receipt and acceptance should be inserted later Abstract
We are developing lumped-element kinetic inductance detectors (LEKIDs) de-signed to achieve background-limited sensitivity for far-infrared (FIR) spectroscopy on astratospheric balloon. The Spectroscopic Terahertz Airborne Receiver for Far-InfraRed Ex-ploration (STARFIRE) will study the evolution of dusty galaxies with observations of the[CII] 158 µ m and other atomic fine-structure transitions at z = . − .
5, both through di- rect observations of individual luminous infrared galaxies, and in blind surveys using thetechnique of line intensity mapping. The spectrometer will require large format ( ∼ × − W Hz − / .The low-volume LEKIDs are fabricated with a single layer of aluminum (20 nm thick) de-posited on a crystalline silicon wafer, with resonance frequencies of 100 −
250 MHz. Theinductor is a single meander with a linewidth of 0.4 µ m, patterned in a grid to absorb op-tical power in both polarizations. The meander is coupled to a circular waveguide, fed by aconical feedhorn. Initial testing of a small array prototype has demonstrated good yield, anda median NEP of 4 × − W Hz − / . Keywords
Kinetic Inductance Detector, Aluminum, Far-Infrared Spectroscopy, Balloon STARFIRE
Instrument
Understanding the formation and evolution of galaxies is one of the foremost goals of astro-physics and cosmology today. The cosmic star formation rate rose dramatically from earlytimes to a peak at approximately half the present age of the universe (at redshift z ∼ STARFIRE (the Spectroscopic Terahertz Airborne Receiver for Far-InfraRed Exploration) is
1: California Institute of Technology, Mail Code 301-17, 1200 E. California Blvd., Pasadena, CA 91125,USA; E-mail: [email protected]: University of Pennsylvania Department of Physics & Astronomy, 209 S 33rd St., Philadelphia, PA, 19104,USA3: Jet Propulsion Laboratory, 4800 Oak Grove Drive, Pasadena, CA, 91109, USA4: National Institute of Standards and Technology, 325 Broadway, Boulder, CO, 803055: Fermi National Accelerator Laboratory, PO Box 500, Batavia IL 60510 a r X i v : . [ a s t r o - ph . I M ] A p r S. Hailey-Dunsheath et al. Fig. 1:
Electromagnetic simulation of the cou-pling efficiency of a TE mode from the circularwaveguide into the aluminum absorber, for sheetimpedances of 2 Ω / (cid:3) ( purple ) and 1 Ω / (cid:3) ( red ),and a 27 µ m backshort. Half power frequencies ofthe testbed bandpass filter are also shown ( dashed ).The band-averaged coupling efficiencies are 94%and 88% for 2 Ω / (cid:3) and 1 Ω / (cid:3) , respectively, anddrop to 22% and 20% with no backshort. (Colorfigure online.) designed to study the interstellar medium (ISM) of galaxies from 0 . < z < .
5, primarilyin the [C II ] 158 µ m line, and also in cross-correlation with the [N II ] 122 µ m transition. STARFIRE will be capable of making a high significance measurement of the [C II ] powerspectrum in at least 4 redshift bins, and of measuring the [C II ] × [N II ] power spectrum at z ∼ STARFIRE will also be able to detect emission lines in a blind survey, and bycorrelating with known optical galaxies measure the [C II ] luminosity of this population aswell. STARFIRE is able to achieve its substantial increase in performance – better than the air-borne instruments on SOFIA or the space-borne
Herschel -SPIRE FTS – by using dispersivespectroscopy to lower the photon noise per detector, and by taking advantage of the consid- erably lower atmospheric background at balloon rather than aircraft altitudes.
STARFIRE willfield two large format ( ∼ . × − W Hz − / . STARFIRE serves as atechnology advancement platform for the Origins Space Telescope [9], and detector devel-opment is currently funded by NASA [2].
STARFIRE will deploy arrays of kinetic inductance detectors (KIDs) with the same single-layer architecture developed for the MAKO camera [8]. A single 20 nm thick aluminumlayer forms both the inductor and interdigitated capacitor, which are designed to achieve100 −
250 MHz readout frequencies. To maximize responsivity the absorber is low volume( V = µ m ), and couples to incident radiation with a circular waveguide fed with a conicalfeedhorn. The waveguide design includes a flare at the bottom of the waveguide and a litho-graphically patterned choke structure to help eliminate conversion into substrate modes. Thefinal pixel will be fabricated on a SOI wafer and will have a 27 µ m thick backshort, createdby etching from the backside to a buried oxide layer, then depositing gold. Electromagneticsimulations indicate band-averaged coupling efficiencies of ≈
90% with the backshort inplace, and ≈
20% without (Figure 1). Initial testing of a device with the backshort fabricatedindicate the presence of the gold has no noticeable impact on the resonator Q .The inductor/absorber is a single meander of 0.4 µ m wide aluminum, patterned to pro-vide an optimal impedance match to the waveguide. The meander effectively couples as amesh to both polarizations by allowing the various segments of meander line to come closeenough to one another at the corners to create capacitive shorts at the optical frequencies.This is achieved with a 0.3 µ m gap and a 0.6 µ m overlap length for each of the intersections(Figure 2). evelopment of Aluminum LEKIDs for Balloon-Borne Far-IR Spectroscopy 3 Fig. 2: (A) Diagram of the mask layout for a single resonator. The meandered inductor ( green ) is sur-rounded by an optical choke structure ( blue ). An interdigitated capacitor ( red ) sets the resonance frequencyof the pixel, and two coupling capacitors ( yellow ) couple microwave signal onto microstrip feedlines. (B) Amicroscope image of a single pixel as fabricated. All pixel elements of the prototype array are patterned outof 40 nm Al film. (C) A microscope image of the 45-pixel prototype array, as fabricated. (D) The fabricatedarray in its enclosure. The back side of the die is bare silicon, and lies flat on the gold-plated package surface.The full size of the die is 30mm × We have cryogenically tested a 45-pixel prototype detector array, fabricated in the JPL Mi-crodevices Laboratory. To measure the performance of detectors with different film thick-nesses we fabricated wafers with both 20 nm and 40 nm thick aluminum, and conducted theinitial testing using an array with a 40 nm thick film. This results in a larger inductor volume( V = µ m ) and higher resonance frequencies than expected for the full STARFIRE array,which will use a 20 nm thick film. Measurements of the sheet impedances of these filmsare in preparation. The array is cooled by a He sorption fridge to a base temperature of210 mK, and exposed to a cryogenic blackbody for optical testing. We use two metal-meshfilters mounted on the detector package to define the optical band: a bandpass filter transmit-ting over ≈ −
900 GHz, and a 1000 GHz cutoff low-pass filter. We use a ROACH-basedreadout system originally developed for use with MAKO [8]. The prototype device had atotal yield of 89% (40/45 resonators), but we focus our analysis on the 11 resonators withresonances below 250 MHz.3.1 Dark MeasurementsOur first step is to characterize the detectors dark, with the feedhorns blanked off. We mea-sure the resonator frequency, Q , and noise as a function of stage temperature. We modelour resonators following the standard application of Mattis-Bardeen theory, along with the S. Hailey-Dunsheath et al. Fig. 3:
Fractional frequency shift and Q − r vs. T stage for a dark measurement, along with fits for α and T c .Models are x = x MB + δ x for a zero temperature offset δ x , and Q − r = Q − + Q − , with Q − a fixed termthat includes the resonator coupling Q and a limiting inductor internal Q . (Color figure online.) assumption that the quasiparticle lifetime depends on the quasiparticle number density as τ qp = τ max ( + n qp / n ∗ ) − , for constants τ max and n ∗ [5, 11]. For a general temperature andabsorbed optical power ( P abs ), n qp may be written as: n qp = − n ∗ + (cid:20) ( n ∗ + n th ) + n ∗ η pb P abs τ max ∆ V (cid:21) . , (1)where n th = N √ π k B T ∆ exp ( − ∆ / k B T ) is the quasiparticle density in thermal equilib-rium, η pb is the pair-breaking efficiency, ∆ = . k B T c is the gap energy, V is the inductorvolume, and we adopt a density of states of N = . × µ m − eV − [5]. The fractionalfrequency shift and the internal Q of the inductor are then written as: x MB = − αγ S N ∆ n qp and Q − = αγ S N ∆ n qp , (2)where γ = S and S [11], but in place of the physical temperature of our devices we substitutean effective electron temperature, obtained by inverting n th ( T ) for n qp [3]. This modifiedapproach becomes important when the absorbed power is nonzero, and n qp > n th . In Figure3 we show measurements of x and Q − r as a function of T stage for one resonator, along withfits for α and T c . The resonators are well-characterized by Equations 1 and 2, with medianvalues of α = .
74 and T c = .
39 K.We measure the fractional frequency noise ( S xx ) at each stage temperature, driving theresonators ∼ S xx (Figure 4). Our general model for the white noise in our KIDs com-bines photon generation noise, thermal generation noise, and recombination (of all quasipar-ticles) noise, along with a fixed term ( S xx , ) representing additional noise sources, assumedto be independent of temperature and optical loading: S xx = (cid:18) αγ S N ∆ (cid:19) (cid:20)(cid:18) η pb τ qp ∆ V (cid:19) h ν P abs ( + n γ ) + ( τ qp ) V ( Γ th + Γ r ) (cid:21) + S xx , , (3) evelopment of Aluminum LEKIDs for Balloon-Borne Far-IR Spectroscopy 5 Fig. 4:
Left
Amplifier-subtracted S xx for a representative KID measured dark ( black ), and with the cryo-genic blackbody delivering P abs =
30 fW ( blue ) and 270 fW ( red ). White noise is estimated by averagingover 30 −
80 Hz ( cyan ), and conversion to a detector NEP uses the responsivity R = . × W − mea-sured at the lowest optical loading P abs =
20 fW.
Right
Amplifier-subtracted S xx measured dark at variousstage temperatures, averaged over 6 KIDs. Fits to Equation 4 are shown with S xx , = red ) and S xx , a freeparameter ( blue solid ), with the latter decomposed into thermal GR ( blue dotted ) and fixed noise floor ( bluedashed ). (Color figure online.) where n γ is the photon occupation number in the detector, the thermal generation rate is Γ th =( n th V / )( τ − + τ − ) , τ th is the quasiparticle lifetime when n qp = n th , and the recombination rate is Γ r = ( n qp V / )( τ − + τ − ) [11]. In the limit of no optical loading this becomes: S xx → (cid:18) αγ S N ∆ (cid:19) n th τ th V (cid:18) + τ th τ max (cid:19) + S xx , . (4)The temperature dependence of Equation 4 is dominated by the n th τ th ( + τ th / τ max ) product.At low temperatures this term reduces to 2 n th τ max and increases rapidly with temperature,while at high temperatures this term asymptotes to n ∗ τ max [6]. In Figure 4 we show the whitenoise as a function of temperature, averaged over a subset (6/11) of the KIDs with the lowestamplifier noise. We show a fit to the data with fixed P abs = S xx , =
0, and n ∗ and τ max as freeparameters. This fit is poor, but the data are well reproduced by introducing a noise floor S xx , ≈ . × − Hz − , or alternatively by assuming a frequency-dependent stray power P abs ≈ (
100 GHz / ν ) fW for a minimum pair-breaking frequency of 100 GHz. These fitsindicate thermal generation-recombination (GR) noise is subdominant to other noise sourcesat our 210 mK operating temperature, but dominates at T >
250 mK.3.2 Optical MeasurementsIn a second test we expose the detector package to the cryogenic blackbody, and measure thefrequency response and noise as a function of the blackbody temperature for T BB = . − η opt ): δ x δ P inc = δ x δ P abs η opt = αγ S N ∆ η pb τ qp ∆ V η opt , (5) S. Hailey-Dunsheath et al. Fig. 5:
Left
Fractional frequency shift vs. P inc with model fit ( red ). Right S xx vs. P inc with model fit ( red ),decomposed into contributions from photon generation noise ( blue solid ), photon recombination noise ( bluedashed ), thermal GR noise ( magenta ), and fixed noise floor ( green ). Vertical dashed lines in both figuresmark the minimum and median STARFIRE optical loading of 120 fW and 200 fW, respectively. (Color figureonline.) and the flattening of this curve at high P inc is consistent with the expected decrease in τ qp at large n qp .We see an increase in white noise with P inc that we attribute to photon noise (Figure 4).The contribution of photon noise to S xx is: S xx , γ = (cid:18) δ x δ P inc (cid:19) h ν P inc η opt (cid:18) + ∆ h νη pb (cid:19) , (6)where we have neglected wave noise (negligible for T BB ≤
11 K), and the last term accountsfor the recombination noise associated with optically-generated quasiparticles. With an as-sumption of η pb = .
57 this term is 1.21, and Equation 6 then demonstrates how the ratio of S xx , γ and ( δ x / δ P inc ) P inc may be used to estimate η opt .We fit the response and noise data shown in Figure 5 using our full resonator model(Equations 1 − n ∗ , τ max , η opt , and S xx , . The data are well fit withthis model, and we find median values of n ∗ = µ m − , τ max = µ s, η opt = .
17, and S xx , = . × − Hz − . This value of τ max is shorter than reported by other groups foraluminum [7, 4]. Our aluminum is currently deposited through a sputter deposition tech-nique; a future shift to electron beam evaporation may produce higher purity films andlonger quasiparticle lifetime, increasing the responsivity [1]. This optical efficiency is closeto the 0 . − .
22 range estimated from simulations of the coupling efficiency between thewaveguide and absorber (Figure 1). The inferred responsivity at our lowest optical loading( P abs =
20 fW) averaged over the array is δ x / δ P abs = . × W − . We combine this withour dark noise measurements to obtain detector NEPs, finding a median value of 4 × − W Hz − / (Figure 4).The STARFIRE arrays will operate with a typical optical loading of 200 fW, at a 250mK base temperature. Increasing the optical load from 20 fW to 200 fW decreases theresponsivity in our prototype KIDs by ≈
40% (Figure 5). Additionally, the dark noise at 250mK is ≈ . × − W Hz − / with the optical load and operating temperature envisioned for STARFIRE ,less than the photon NEP of 1 . × − W Hz − / . evelopment of Aluminum LEKIDs for Balloon-Borne Far-IR Spectroscopy 7 We have fabricated and characterized a 45-pixel
STARFIRE prototype detector array. TheseLEKIDs are low volume (76 µ m ) devices fabricated with a single layer of 40 nm thickaluminum, are sensitive to both polarizations, and couple to free space with circular waveg-uide and conical feedhorns. Operating at 210 mK we measure a typical NEP of 4 × − W Hz − / , and confirm that thermal GR noise is not the dominant noise source. With a 250mK operating temperature and under a 200 fW load, as we anticipate for STARFIRE , thedetector NEP will be ≈ × − W Hz − / . This compares favorably to the typical photonNEP of 1 . × − W Hz − / . Acknowledgements
ACMB’s work was supported by a NASA Space Technology Research Fellowship.Detector development for
STARFIRE is supported by NASA grant 15-APRA15-0081. We thank C. Groppi forgenerously providing the tool used to drill the feedhorns.
References
1. R. Barends, S. van Vliet, J. J. A. Baselmans, S. J. C. Yates, J. R. Gao, and T. M. Klap-wijk. Enhancement of quasiparticle recombination in Ta and Al superconductors by implantation of magnetic and nonmagnetic atoms.
Phys. Rev. B , 79(2):020509, Jan.2009. doi: 10.1103/PhysRevB.79.020509.2. A. Barlis, S. Hailey-Dunsheath, C. M. Bradford, C. McKenney, H. G. Le Duc, andJ. Aguirre. Development of low-noise kinetic inductance detectors for far-infrared as-trophysics. In
APS April Meeting Abstracts , Jan. 2017.3. P. Barry. J. Low Temp. Phys. In
This Special Issue , 2017.4. J. J. A. Baselmans, J. Bueno, S. J. C. Yates, O. Yurduseven, N. Llombart, K. Karatsu,A. M. Baryshev, L. Ferrari, A. Endo, D. J. Thoen, P. J. de Visser, R. M. J. Janssen,V. Murugesan, E. F. C. Driessen, G. Coiffard, J. Martin-Pintado, P. Hargrave, andM. Griffin. A kilo-pixel imaging system for future space based far-infrared observa-tories using microwave kinetic inductance detectors.
A&A , 601:A89, May 2017. doi:10.1051/0004-6361/201629653.5. J. Gao, J. Zmuidzinas, A. Vayonakis, P. Day, B. Mazin, and H. Leduc. Equivalence ofthe Effects on the Complex Conductivity of Superconductor due to Temperature Changeand External Pair Breaking.
Journal of Low Temperature Physics , 151:557–563, Apr.2008. doi: 10.1007/s10909-007-9688-z.6. P. D. Mauskopf, S. Doyle, P. Barry, S. Rowe, A. Bidead, P. A. R. Ade, C. Tucker,E. Castillo, A. Monfardini, J. Goupy, and M. Calvo. Photon-Noise Limited Performancein Aluminum LEKIDs.
Journal of Low Temperature Physics , 176:545–552, Aug. 2014.doi: 10.1007/s10909-013-1069-1.7. H. McCarrick, D. Flanigan, G. Jones, B. R. Johnson, P. Ade, D. Araujo, K. Brad-ford, R. Cantor, G. Che, P. Day, S. Doyle, H. Leduc, M. Limon, V. Luu,P. Mauskopf, A. Miller, T. Mroczkowski, C. Tucker, and J. Zmuidzinas. Horn-coupled,commercially-fabricated aluminum lumped-element kinetic inductance detectors formillimeter wavelengths.
Review of Scientific Instruments , 85(12):123117, Dec. 2014.doi: 10.1063/1.4903855.8. C. M. McKenney, H. G. Leduc, L. J. Swenson, P. K. Day, B. H. Eom, and J. Zmuidzinas.Design considerations for a background limited 350 micron pixel array using lumpedelement superconducting microresonators. In
Society of Photo-Optical Instrumentation
S. Hailey-Dunsheath et al. Engineers (SPIE) Conference Series , volume 8452 of
Society of Photo-Optical Instru-mentation Engineers (SPIE) Conference Series , Sept. 2012. doi: 10.1117/12.925759.9. M. Meixner, A. Cooray, R. Carter, M. DiPirro, A. Flores, D. Leisawitz, L. Ar-mus, C. Battersby, E. Bergin, C. M. Bradford, K. Ennico, G. J. Melnick, S. Milam,D. Narayanan, K. Pontoppidan, A. Pope, T. Roellig, K. Sandstrom, K. Y. L. Su, J. Vieira,E. Wright, J. Zmuidzinas, S. Alato, S. Carey, M. Gerin, F. Helmich, K. Menten, D. Scott,I. Sakon, and R. Vavrek. The Far-Infrared Surveyor Mission study: paper I, the genesis.In
Space Telescopes and Instrumentation 2016: Optical, Infrared, and Millimeter Wave ,volume 9904 of
Society of Photo-Optical Instrumentation Engineers (SPIE) ConferenceSeries , page 99040K, July 2016. doi: 10.1117/12.2240456.10. B. D. Uzgil, J. E. Aguirre, C. M. Bradford, and A. Lidz. Measuring Galaxy Clusteringand the Evolution of [C II] Mean Intensity with Far-IR Line Intensity Mapping during0.5 < z < ApJ , 793:116, Oct. 2014. doi: 10.1088/0004-637X/793/2/116.11. J. Zmuidzinas. Superconducting microresonators: Physics and applications.
An-nual Review of Condensed Matter Physics , 3(1):169–214, 2012. doi: 10.1146/annurev-conmatphys-020911-125022. URL http://dx.doi.org/10.1146/annurev-conmatphys-020911-125022http://dx.doi.org/10.1146/annurev-conmatphys-020911-125022